What are some hurdles pharmaceutical and healthcare organizations need to navigate to succeed with AI?
Ng: One key hurdle is the lack of a unified strategy. A recent McKinsey & Company survey found that about 75% of respondents say their organizations lack a comprehensive vision for gen AI or a strategic roadmap with clearly defined success measures tied to business priorities. Organizations must address this by implementing an enterprise-wide vision, as without it, it is difficult to sustain momentum or scale promising pilots.
Another challenge lies in data fragmentation. While pharmaceutical companies generate massive volumes of data from clinical trials, much of it remains siloed. Without a unified data foundation, AI models struggle to generate actionable insights, ultimately slowing innovation and decision-making.
To overcome this, organizations should consider investing in integrated, end-to-end platforms that connect data across the entire value chain. Medidata’s end-to-end unified platform, for example, manages the full clinical trial lifecycle from study design and patient recruitment to monitoring, data analysis, and results reporting. Such unified approach helps break down data silos and creates a consistent, high-quality dataset that supports more effective use of AI across the board.
Risk management is also a critical area that is often under-addressed. The McKinsey survey found that 35% of respondents spend fewer than 10 hours collaborating with risk counterparts. This signals that risk mitigation is not sufficiently embedded in AI development processes, which is especially important now as we see gen AI introducing novel risks, including hallucinations, bias, IP exposure, and data privacy issues.
At Medidata, we believe that these risks must be proactively managed across the entire lifecycle, not just treated as compliance checkpoints. We are focused on leveraging AI across every stage of the clinical trial process including the identification and management of potential risks. This includes integrating AI analytics that enable real-time risk monitoring, as well as developing pioneering innovations like Medidata Simulants, which uses gen AI to produce high-fidelity synthetic data based on a database of over 36,000 clinical trials, and helps to uncover potential risks early.
What’s key though is that while genAI can support decision-making, it should not replace human judgment. Guardrails are essential, including human review of AI-generated content, especially when it influences treatment decisions or patient communication.
How would a fully AI-integrated clinical workflow look like, and what regulatory and operational shifts are needed?
Ng: A fully AI-integrated clinical workflow would embed intelligent, data-driven decision-making across every stage of drug development. AI does not simply enhance isolated processes, but it transforms the entire ecosystem. Specifically, AI would play a strategic role across:
- Trial design: AI enables organizations to make more informed decisions when selecting and designing clinical development programs by simulating different trial scenarios and identifying the most effective study designs.
- Planning & execution: It helps accelerate trial timelines by optimizing site selection. AI can also reduce operational costs and enhance quality through better recruitment planning and real-time risk management.
- Regulatory filing: AI improves the way data is organized and analyzed, allowing sponsors to better demonstrate a treatment’s value to regulators, payers, and patients through clearer, more comprehensive evidence.
- Commercialization & medical use: It strengthens the clinical evidence base and supports personalized treatment strategies, helping ensure that the right therapies reach the right patients at the right time.
AI integration is also essential in decentralized clinical trials (DCTs), which depend heavily on digital technologies like remote monitoring, electronic informed consent (eConsent), and electronic clinical outcome assessments (eCOA). These tools generate complex, real-time data, and AI supports the management of this data. It can further ensure data quality and integrity throughout the trial.
AI’s role will expand beyond research operations and into patient relationships. By 2040, we anticipate a landscape where AI is woven throughout the healthcare journey, which enables personalized treatments, seamless data sharing, and more responsive care.
This shift will require reimagining not only the technical architecture of trials but also the regulatory frameworks that govern them.
To enable this future, regulators and industry sponsors must collaborate closely to validate and evolve new AI-driven methodologies. Flexibility must go hand-in-hand with scientific rigor to maintain trust and safety, and there is also a need to ensure inclusivity. With intuitive interfaces and adequate support, even patients less familiar with digital tools can participate confidently.
The goal should be to design trials that adapt to patients’ lives and not the other way around.
